Educational Evaluation and Policy Analysismkennedy/publications/docs/Teacher... · Contributions of Qualitative Research to Research on Teacher Qualifications Mary M. Kennedy Michigan

http://eepa.aera.net

Analysis Educational Evaluation and Policy

DOI: 10.3102/0162373708326031 2008; 30; 344 EDUCATIONAL EVALUATION AND POLICY ANALYSIS

Mary M. Kennedy Contributions of Qualitative Research to Research on Teacher Qualifications

http://epa.sagepub.com/cgi/content/abstract/30/4/344 The online version of this article can be found at:

Published on behalf of

http://www.aera.net

By

http://www.sagepublications.com

can be found at:Educational Evaluation and Policy Analysis Additional services and information for

http://eepa.aera.net/cgi/alerts Email Alerts:

http://eepa.aera.net/subscriptions Subscriptions:

http://www.aera.net/reprintsReprints:

http://www.aera.net/permissionsPermissions:

at MICHIGAN STATE UNIV LIBRARIES on December 18, 2008 http://eepa.aera.netDownloaded from

http://www.aera.net

http://eepa.aera.net/cgi/alerts

http://eepa.aera.net/subscriptions

http://www.aera.net/reprints

http://www.aera.net/permissions

http://epa.sagepub.com

344

Contributions of Qualitative Research to Research on Teacher Qualifications

Mary M. KennedyMichigan State University

The influence of teachers’ qualifications on their teaching practice has been subject to debate.Literature reviews do not settle these debates, partly because the literature is uneven and partlybecause reviews capture only narrow slices of literature. In particular, many reviews eliminate qual-itative studies. Yet without examining qualitative evidence, variations in quantitative findings aredifficult to interpret, disappointing findings are difficult to understand, and plausible explanations ofpatterns are in short supply. The present review focuses on qualitative studies and compares theirfindings with those from quantitative literature. The author finds that the qualitative literature agreeswith quantitative literature in its inability to distinguish between teachers with different types ofcertificates or different teacher education backgrounds. On the other side, the author finds moreevidence of benefit from content knowledge than quantitative studies have typically found. The quali-tative studies also reveal competing influences and offer hypotheses about why the outcomes look theway they do.

Keywords: teacher qualifications, qualitative methods, teacher education

THE volume of research on teachers’ qualifica-tions has grown to several hundred studies, butit has not settled arguments about the merits ofteacher education programs. One reason for thisstalemate is that the question is difficult toaddress, and virtually all studies are susceptibleto confounding variables. For instance, teachersself-select their educational programs; conse-quently, their own values and predispositionsare confounded with their credentials. And oncecertified, they engage in nonrandom job-seekingpractices, while districts engage in nonrandomhiring practices, so the resulting pattern of jobplacements confounds teachers’ educational

backgrounds, credentials, attitudes, and predis-positions with the communities they serve(Boyd, Lankford, Loeb, & Wykoff, 2003;Lankford, Loeb, & Wykoff, 2002; Strauss,1999). Kennedy, Ahn, and Choi (2008) call theresulting allocation of teachers to schools a sys-tem of affinity assignments, meaning that teach-ers’ social backgrounds and qualifications oftenmatch the social backgrounds and “qualifica-tions” of their students.

The problem is compounded when literaturereviews use different inclusion rules. Somereviewers (e.g., R. Greenwald, Hedges, &Laine, 1996; Wayne & Youngs, 2003; Wilson,

The author acknowledges financial support from the U.S. Department of Education’s Office of Educational Research andImprovement, now Institute for Educational Sciences; by the National Science Foundation’s Division of Research, Evaluation,and Communication; and by the National Science Foundation’s Math and Science Partnerships. However, the opinionsexpressed here are the author’s, and no institutional endorsement should be inferred.

Educational Evaluation and Policy AnalysisDecember 2008, Vol. 30, No. 4, pp. 344–367

DOI: 10.3102/0162373708326031© 2008 AERA. http://eepa.aera.net



345

Research on Teacher Qualifications

Floden, and Ferrini-Mundy, 2001) do not reachbeyond journal articles, whereas others cast abroader net. Some consider only studies that usestudent achievement as their outcome measure(e.g., R. Greenwald et al., 1996; Wayne &Youngs, 2003), whereas others include an arrayof indicators of quality teaching (Rice, 2003;Wilson et al., 2001). Some (e.g., Kennedy et al.,2008) require that outcomes be measured afterteachers have finished their education and areemployed as teachers of record, whereas others(e.g., Wilson et al., 2001) include studies whoseoutcomes were measured while teachers werestill participating in student teaching. Some areunclear about their inclusion rules (e.g., Walsh,2001). Finally, each review also imposes its owncriteria for study quality. The number of studiesincluded in reviews of literature on teacher qual-ifications ranges from 21 in Wayne and Youngs’s(2003) review to 200 in Walsh’s (2001) review.

In part because different readers of this liter-ature focus on different subsets within it, wealso can see very different conclusions arisingas different authors perceive the available evi-dence in very different ways. Here are someillustrative quotes:

More than 200 studies have found that teachers whohave more background in their content areas and havegreater knowledge of teaching and learning are morehighly rated and more successful with students infields ranging from early childhood and elementaryeducation to mathematics, science and vocationaleducation. (Darling-Hammond & McLaughlin, 1999,pp. 377–378)

I would conclude, if anything, that there is little sup-port for the impact of teachers’ subject matter knowl-edge on student learning. (Friedman, 2000, p. 20)

Contrary to conventional wisdom, mathematics andscience students who have teachers with emergencycredentials do no worse than students whose teach-ers have standard teaching credentials, all else beingequal. (Goldhaber & Brewer 1999, p. 97)

The evidence indicates that neither an extreme cen-tralized bureaucratization nor a complete deregula-tion of teacher requirements is a wise approach forimproving teacher quality. (Rice, 2003, p. vi)

These quotes reveal that the relationship betweenteacher qualifications and teaching quality—arelationship that should be self-evident—is notself-evident at all. Moreover, the field lacks

coherent theoretical arguments about howqualifications are expected to influence teachingpractice. This is an area, therefore, where quali-tative research could make a strong contribu-tion, for qualitative research is more able todelve into the mechanisms and processes bywhich these qualifications actually influenceteaching practice.

My contribution to this debate is to add qualita-tive studies to the corpus of literature under review.Because these studies have rarely been consideredin reviews, I also offer a framework for evaluatingthe validity of their causal claims. And to gauge theimport of their contribution, I contrast their con-clusions with those from other literature reviews,and I ask whether they help fill the theoreticalgap regarding how knowledge and educationalbackgrounds influence teaching practice.

Most reviews exclude qualitative research,either because its outcomes are not quantifiableor because of a perception that qualitative stud-ies are useful for learning about social meaningsbut not about causal influences. Yet many quali-tative studies do examine the influence of poli-cies or programs on teaching practice, and theycan help us understand these influences in away that quantitative studies cannot. Maxwell(2004a, 2004b) has argued that qualitative stud-ies address causation not by establishing pat-terns of regularities, as quantitative studies do,but by revealing the mechanisms and processesof influence. In the case of teaching, qualitativestudies can help us see how a teacher’s knowl-edge either misdirects practice or enhances itand see where and how educational back-grounds have their influences, if they do.Qualitative research is also the preferred methodof teacher educators, as evidenced by the largefraction of qualitative studies in journals suchas the Journal of Teacher Education, so whenreviewers exclude this literature, they areexcluding the bulk of work that has been doneby teacher educators themselves.

Qualitative research has been a staple of socialsciences for some time but ascended in educa-tional research in the 1980s. Since then,researchers have struggled to articulate the dif-ferences between quantitative and qualitativeresearch and to articulate standards of validity forqualitative research. It has often been referred to



346

Kennedy

not merely as an alternative method but as analternative paradigm, with different assump-tions, methods, rationale, and purposes. Somereviewers (e.g., Wilson et al., 2001) label thiswork “interpretive,” suggesting that the natureof one’s data automatically implies an alterna-tive epistemology. Others (e.g., Becker, 1996)insist that although qualitative and quantitativeapproaches are different, their epistemologiesare not. Maxwell (2004a, 2004b) argued that thetwo approaches are subject to many of the sameconcerns regarding validity but that the formerrelies on variance and regularities to establishcausality, whereas the latter relies on revealingthe causal mechanisms themselves. Donmoyer(2001) argues that there is no single way tocharacterize qualitative studies, for they varysubstantially in their aims. He suggests that theybe defined by their purpose rather than by theirepistemologies or paradigms. Included in his listof purposes is one called “truth-seeking.” Truth-seeking qualitative studies differ from thosewhose purpose is, say, to create a personal inter-pretation and those whose purpose is socialchange. The fact that meta-analysts are nowseeking approaches to incorporating qualitativestudies into their work (see, e.g., Hundersmarck,2004; Au, 2007) suggests a growing interest inthe truth-seeking potential of qualitative studies.

The central question guiding the presentreview is whether and how truth-seeking quali-tative studies can amend, clarify, or otherwiseadd to the knowledge gained from the largerquantitative body of literature regarding theinfluence of teachers’ qualifications on the qual-ity of their teaching practices. All of the studiesreviewed here examine the influence of teach-ers’knowledge or educational backgrounds on thequality of their teaching practice, and all seemedto expect their findings to be generally relevant.In this sense, they fit Donmoyer’s (2001) cate-gory of truth-seeking studies.

The article has three main sections. In thefirst, I describe the literature search process. Inthe second, I examine methodological issuesassociated with truth-seeking qualitative studiesand offer some criteria for evaluating theirvalidity. In the third, I review the findings fromthese studies. I organize the findings around dif-ferent kinds of qualifications.

Search Procedures

This review is based on literature in theTeacher Qualifications and the Quality ofTeaching (TQQT) database. This database con-tains a collection of research articles, reports,dissertations, conference presentations, andbooks that examine the relationship between atleast one teacher qualification and at least oneindicator of the quality of teaching, with qual-ifications defined as aspects of education orcertification that have been the target of educa-tion policy. Literature in ERIC, PsycInfo, andEconLit was searched using more than 20terms to characterize teachers’ educationalbackgrounds: teacher knowledge, test scores,preparation, education, courses taken, and soforth. The search also included more than 20terms to characterize evidence of teachingquality: authentic practice, quality of practice,student achievement, classroom practice,classroom activities, and so forth. Studies hadto be published after 1960 and had to investi-gate K–12 teachers of record in the UnitedStates. Studies of student teachers were elimi-nated, as were those that placed teachers inartificial settings or in simulations. This lastrequirement eliminated a number of small-scale experimental studies.

Each study also had to establish a linkbetween the qualification and the quality ofpractice, rather than simply describing both andasserting that they were related. Here are twoexamples of qualitative studies that wererejected for failure to establish a link. In onecase (Pool, Ellett, Schiavone, & Carey-Lewis2001), the authors purported to test the linkbetween National Board Certification and teach-ing quality. They drew a sample of six board-certified teachers, interviewed them, andobserved their practices. They found that theteachers’ practices were quite variable andargued that this variability constituted evidencethat board certification was not associated withteaching quality. We rejected the argument thatvariability within a group was, by itself, evi-dence that group was not distinguished relativeto the population as a whole. In another study,Wozniak (1990) set out to learn the backgroundsof a sample of K–12 art teachers who had been



identified as outstanding. The author thenreported common themes and argued that thecharacteristics she found must have beenresponsible for the quality of teachers’ practice.Nothing in the study showed readers how theseteacher characteristics actually fostered teach-ing quality, nor did the author examine a com-parison group of less outstanding teachers to seeif they lacked these qualifications or had them.Notice that in both of these cases, studies wererejected because of the logic of their claim, notthe quality of the evidence.

The results of this search-and-screen processwas a database of more than 450 study reports,including 23 studies describing findings thatwere primarily qualitative but that spoke to thecausal question of whether or how teachers’knowledge or educational backgrounds influ-enced the quality of their teaching practice.Below, I discuss methodological issues uniquelyassociated with these 23 qualitative studies.

Defining Validity in Truth-SeekingQualitative Research

A great deal of ink has been spilled in thecause of defining validity for qualitativeresearch, and my aim is not to review all of thathere. Instead, I concern myself with validityissues associated with a narrow slice of qualita-tive research that is truth seeking in its focusand that specifically addresses the influence ofteachers’ knowledge and educational back-grounds on the quality of their teaching practices.For this group of studies, criteria for validityare very similar to those faced by quantitativeresearchers: Studies must show evidence that aninfluence actually occurred or failed to occur,they must persuade readers that the observedteachers are not idiosyncratic in some way, andthey must acknowledge, and try to eliminate,any alternative explanations of their findings. Iaddress each of these issues below, beginningwith the alternative explanations that must beovercome and then reviewing study designsand analytic strategies that these researchersemployed. The issues I enumerate below arenot intended to apply to all truth-seeking quali-tative research but to the specific studies in thisdatabase.

Alternative Explanations for Qualitative Findings

Whether quantitative or qualitative, researchon teacher qualifications is virtually all non-experimental. No researcher has randomlyassigned prospective teachers to different kindsof certificates, majors, grade point averages, ortest scores nor assigned them to different schooldistricts or schools. Therefore, all of these stud-ies must be examined for alternative explana-tions for observed patterns in outcomes. Thesealternatives threaten the validity of the inferencethat the qualification of interest was responsiblefor observed outcomes. Here is a list of alterna-tive hypotheses that most often presented them-selves in this set of 23 studies, listed in order ofhow frequently they appeared.

Accommodation bias. One explanation forobserved practice is that teachers are trying toaccommodate the researcher’s interests andvalues. Because qualitative researchers havenumerous interactions with teachers, they havemany opportunities to share their own valuesand beliefs and consequently to “tell” teacherswhat kind of practices they want to see orwhat values and beliefs they want to hear about.The problem is especially severe when theresearcher is a teacher educator studying teach-ers who were his or her students a few years ear-lier. In these cases, teachers would be wellaware of pedagogical theories and practices theresearcher advocated. I do not mean to suggestthat teachers would invent completely phonypersonalities; they could simply emphasizesome aspects of their work more than others,use the researchers’ language to describe theirviews and practices, or make an effort to gener-ate examples of how a program had influencedthem, simply to be helpful. These adjustmentsare a standard part of social practice, but quali-tative researchers need to be alert to this possi-bility and develop data collection proceduresthat can correct for this potential bias. Directobservation of practice, rather than reliance oninterviews, is a strong corrective for accommo-dation bias, as teachers would be unlikely togenerate a practice that they did not normallyproduce.

347



guest.venture

Accommodation


Confirmation bias. Observers may be moreinclined to notice events that fit their own apriori hypotheses than things that do not. Forexample, a researcher who believes a tradi-tional teacher preparation program is superiorto an alternative program may notice moreproblems when observing teachers who wereprepared in an alternative program. Confir-mation bias is a widely recognized human dis-position and can occur in research as well as ineveryday life (A. G. Greenwald, Pratkanis,Leippe, & Baumgardner, 1986). It can be con-trolled by ensuring that observers do not knowwhich teachers come from which educationalbackgrounds (a procedure ironically called“blind observation”), though in this collection ofstudies, I found only two in which observers wereblind to the qualifications of the teachers theywere observing. There are also techniques avail-able to researchers who videotape lessons andthen assign “blind” observers to document events.Maxwell (2004b) offers other strategies thatcan help here as well.

Hindsight. One thing that may be unique to thisbody of literature is an evaluative tone in depic-tions of classroom practice. Researchers hold ide-alized visions of what good teaching looks likeand often have videotapes that can be viewed andre-viewed, so their critique is based on repeatedopportunities to reflect on events and on howthings could have or should have been done. Butthe teacher’s practice was generated sponta-neously and did not have the benefit of reflectivehindsight, so the interpretation of practice may beoverly idealist and unfair to the teacher. One pos-sible antidote for hindsight is to accompany cri-tiques of practice with evidence of the teacher’sown rationale for what occurred and the teacher’sown critique of what occurred.

Reactivity. Sometimes the data collection processitself can influence study outcomes. For instance,if a researcher is examining the relationshipbetween teachers’ knowledge of, say, cultural dif-ferences among students and their classroompractices, he or she might begin by asking teach-ers about the issue, thus increasing the teachers’awareness of it and thereby altering their subse-quent practice. Reactivity can be reduced byusing more indirect interview techniques that

pursue particular topics without overtly announc-ing their salience to interviewees.

Misspecified influences. Researchers interestedin the influence of a particular teacher educationprogram sometimes document program contentafter the teachers have graduated, so the pro-gram the researchers saw at the time of the studymight not be the program that their focal teach-ers saw when they were students. The best wayto prevent this is to document programs at thetime prospective teachers are experiencing themrather than later on.

Confounded contexts. Because qualitativeresearchers tend to focus on only a handful ofcases, differences among teaching practicesmay be influenced by idiosyncratic aspects of local contexts rather than by the teachers’degree or other qualification. Perhaps oneteacher teaches college preparatory classeswhereas the other teaches remedial classes, orperhaps one is in a more supportive school orhas more planning time than the other. It is pos-sible that these variations, rather than a differ-ence in qualifications, are responsible for differ-ences in observed practice. Quantitativeresearchers, with their larger samples, hope thatthese variations will cancel each other out sothat the overall influence of a particular qualifi-cation can be found above the din. A benefit ofqualitative research is that it can incorporatelocal contextual influences into the analysis, butresearchers are very uneven in the attention theygive to these issues. In small samples, theseaspects of context need to be incorporated intothe researchers’ causal accounts.

Establishing Influence in the Face ofAlternative Explanations

Studies in this review were all interested inhow a particular qualification—an educationalbackground or a body of content knowledge—influenced teaching practice. Thus a central taskfor researchers is to establish that such an influ-ence did exist and that the appearance of aninfluence cannot be explained by reference tosome other explanation, such as one of thoselisted above. The studies in this review used arelatively small set of strategies for establishing

348

Kennedy



that a qualification had some influence on prac-tice, often relying on one or more of the follow-ing strategies: (a) between-teacher comparisons;(b) within-teacher comparisons; (c) longitudinalstudies, typically following people from theirteacher-education programs into their practice;(d) teacher attributions of influence; and (e)detailed descriptions that demonstrate the influ-ence itself.

Between-teacher comparisons. About half thestudies engaged in between-teacher comparisons,using a comparison-group logic that was analo-gous to quantitative group comparisons. Forexample, Powell (1997) contrasts two secondaryscience teachers, one who had not majored in thesubject she was teaching and one who had anadvanced degree in the subject plus several yearsof applied experience in the field. The two teach-ers participated in the same teacher educationprogram and, upon graduation, taught the samesecondary science courses.

Within-teacher comparisons. Two studies in thisgroup relied on within-teacher comparisons.They contrasted a single teacher’s practice acrosstwo different subjects. If the research questionhas to do with something relevant to the contenttaught, such as the teachers’ subject matterknowledge, this can be a strong research design.Because the same teacher is teaching bothsubjects, a wide range of other aspects of theteacher’s background, attitudes and values,students, and teaching situation are held constant,so observed differences are easier to interpret.

Longitudinal and content-matching studies. Sixstudies employed a technique of followingteachers from their undergraduate experiencesinto their professional teaching assignments tosee if the ideas taught by the programs appearedin their graduates’ practices or rhetoric. There isa greater burden on these researchers to explainhow they selected their samples of teachers, forthe potential for bias is large. These designs alsoare threatened by the potential for accommoda-tion bias, because teachers are aware of programgoals and values.

Teacher attribution of learning. An obvious strat-egy for linking teachers’ educational backgrounds

to their current teaching practice is to ask themwhere they got their ideas. No researcher usedthis as the only method for establishing influence,but several used this in conjunction with otherstrategies. The strategy is somewhat more risky ifteachers know the purpose of the study and wantto please the researcher. However, a goodresearcher will not ask, “Did you get that idea inour XYZ program?” but instead ask, “Where didyou get that idea?”

Detailed descriptions of the influence itself.Although detail is a widely advertised strength ofqualitative research, it is not widely practiced. Ifound only two studies that provided voluminousand detailed descriptions of observed practices toestablish their causal links (Smith, 1989; Stein,Baxter, & Leinhardt, 1990). Both were case stud-ies of individual teachers, and both providednumerous detailed examples. Such descriptionscan overcome the lack of a comparison as well asmany other alternative explanations.

Establishing Representation With a Small Sample

The third challenge facing all researchers isto be clear about how their samples are selectedand what each case is a case of. Teachers may beselected because they participated in differentprograms, participated in the same program butresponded differently, entered a program withdifferent prior beliefs and values, or took teach-ing positions in different contexts or for someother reason. The rationale for both initial selec-tion and for contrasts is a central part of argu-ments about what was learned from the study.Some authors gave details about the particularteacher or teachers they studied, why theyselected them and why they saw these particularteachers as newsworthy; others gave no expla-nation for their samples.

Critics of qualitative research often arguethat these small samples lack value because theyautomatically deny any possibility of represen-tativeness. And some qualitative researchersapparently agree, for instead of making a casefor the representativeness of their samples,they rely instead on Yin’s (1989) argument thatqualitative studies do not generalize to popula-tions but only to theory. This argument does not

349




absolve researchers, however, for readers willnecessarily ask, Who are these teachers andwhere did they come from? Readers need toknow what the researchers saw as important in aprospective research subject, and they need toknow the kind of teaching assignments, schoolenvironments, and students that sampled teacherscontended with. This kind of information canhelp readers make their own judgments about thelogic of any causal conclusions, about potentialsources of bias in the study, and about the theo-retical relevance of the findings as well.

In fact, the most widespread weakness in thestudies reviewed here was a general failure toarticulate how teachers were selected. Four stud-ies gave no indication at all of how cases wereselected, one based selection on geographicproximity, and another stated explicitly that thestudy capitalized on a serendipitous event. Manyof the researchers already knew their sampledteachers because they had taught them, and yetthey failed to say why they selected these gradu-ates rather than some others.

Study Findings

Findings are presented according to the par-ticular type of qualification examined. I foundstudies examining the following: (a) influence ofcontent knowledge that is stipulated on the basisof a credential, (b) influence of content knowl-edge as assessed by the researcher, (c) influenceof particular teacher education courses, and(d) influence of broader patterns of courses orentire teacher education programs.

Influence of Stipulated Content Knowledge

Policy makers have access to a large volumeof quantitative studies examining the relationshipbetween teachers’ content knowledge and theirstudents’ achievement. Content knowledge hasbeen determined by college majors, degrees andcertificates, test scores, and number of coursestaken. Overall, this work has been disappointingin its failure to demonstrate the expected rela-tionship. For example, one recent review (Ahn &Choi, 2004) synthesizing correlational researchin mathematics education found an averagecorrelation of only .11 between teachers’ tested

content knowledge and their students’ achieve-ment scores. Another review (Kennedy et al.,2008) focused on the number of mathematicscourses teachers took in college but still foundeither very small or very inconsistent resultsacross their studies. Many of these studiesfocused on knowledge of mathematics, where thelack of association between teachers’ contentknowledge and their students’ achievementgrowth has been noticed since at least Begle(1972). Some mathematics educators (e.g., Ball,Hill, & Bass, 2005; Ball, Lubienski, & Mewborn,2001) are now hypothesizing that this long his-tory of mediocre findings might occur becauseteacher tests are not tapping the content knowl-edge that is most relevant to teaching. Studiesexamined in the next section consider a variety ofconceptions of content knowledge.

In this section I review three studies thatexamine the role of content knowledge when theknowledge itself is not directly assessed butinstead is stipulated by the presence of a degreeor credential. The first study involves an authorcontrasting his own teaching in the field forwhich he felt prepared and in another fieldwhere he felt less qualified. The second studycompared a teacher with professional workexperience as a scientist with one who had noapplied experience and who had majored in adifferent branch of science. The third comparedtwo secondary English teachers, one of whomwas certified to teach English, the other ofwhom was certified to teach a different subject.These studies are summarized in Table 1.

In the first study, Carlsen (1997) comparedhis own teaching in a subject he knew well(biology) with his teaching in a subject he knewless well (chemistry). I consider this to be stipu-lated knowledge because Carlsen did not pro-vide any evidence of his knowledge other thanto say that he had it. To see how his knowledgeinfluenced his practice, he tape-recorded hislessons so that he could systematically examinethe transcripts. He was particularly interested inthe role that his questions played in controllingconversation and in reinforcing the teachers’authority. He felt that an overreliance on ques-tions could create an inquisition atmosphere inthe classroom and that an overreliance on recallquestions, without attention to rationale, reducedthe authority of the content. In his comparison,

350

Kennedy



351

Carlsen saw that he used remarkably more ques-tions when teaching chemistry than when teach-ing biology and, in particular, that he used morelow-level, recitational questions. He also foundthat his chemistry lessons were much thinner interms of revealing warrants for knowledgeclaims, so his chemistry students learned lessabout how facts are generated and evaluatedthan his biology students learned. He theorizedthat the difference between his questioning ten-dencies in chemistry versus biology occurredbecause he himself was more uncertain of thecontent and therefore relied on questions thattacitly discouraged, rather than encouraged, stu-dent discussion.

The second study (Powell, 1997) contrastedtwo novice teachers who graduated from theauthor’s own teacher education program. Bothhad completed their teaching degrees and hadtaken positions teaching earth sciences to lower-track high school students. They also sharedmany teaching values: They both espoused pro-gressive pedagogies, both viewed science as away of thinking and reasoning that studentsshould experience firsthand, and both wanted tocreate a science curriculum that was embeddedin activities rather than in the textbook.However, Jill’s content expertise was in biolog-ical sciences, not earth sciences, whereas Danhad a master’s degree in geology and had been afield hydrogeologist for 6 years before enteringthe teacher education program.

Powell found that both teachers had diffi-culty implementing their vision. Jill had diffi-culty because of her lack of knowledge of earth

science. Dan, who had greater content knowl-edge and practical experience as well, did notknow how to transform his deeper content knowl-edge into meaningful learning activities. So bothteachers, despite their differences in contentknowledge, ultimately compromised their plansto make science more active and meaningful tostudents and instead relied heavily on the text-book and on recitational teaching.

In the third study, Ringstaff and Sandholtz(2002) compared two secondary English teach-ers, one teaching out of field (Brian, certified inscience) and one certified in English (David).Both graduated from the same university, partic-ipated in the same teacher preparation program,were 1st-year teachers, and received little sup-port from their colleagues. The authors watchedthem teach Cannery Row and found that thepresence of a teaching certificate in English didnot promote better teaching. Brian, the sciencemajor, taught a low-track class but set high goalsfor his students, whereas David, the Englishmajor, taught a college preparation class butdeemphasized literary analysis and focusedinstead on low-level knowledge. The authorsconclude that the official degree is not a perfectindicator of content knowledge, nor of intellec-tual interest in the content, for Brian was an avidreader. Second, they note that participation inteacher education is not a perfect indicator. Bothteachers attended the same teacher preparationprogram, but Brian was generally more able tomanage his students and garner their interest inthe content. Third, time available for planningcould have been a complicating factor, because

TABLE 1Studies of Stipulated Content Knowledge

Teaching Qualification Evidence of quality Strategy for Citation n responsibility of interest of teaching practice establishing link Link established?

Carlsen 1 Secondary College major Questioning Within-teacher Visible influence(1997) science strategies comparison

Powell 2 Secondary Degree and Broad approach Between-teacher Weak influence(1997) science experience to teaching comparison

Ringstaff & 2 Secondary In field vs. Broad approach Between-teacher No visible Sandholtz English out of field to teaching comparison influence but (2002) findings

compromised by confounded context



David had four preparations per day whereasBrian had only two.

Influence of Assessed Content Knowledge

Eight studies directly assessed teachers’ con-tent knowledge and then looked at how that waslinked to their teaching practices. Their strate-gies were remarkably varied, and many incorpo-rated aspects of quantitative research as well asqualitative research. They are summarized inTable 2. For ease of discussion, I sort them bygrade span and content domain.

Elementary mathematics. Three studies examinedthe influence of teachers’ content knowledge onteaching practices in elementary mathematics.The first study examined 20 experienced teacherswho had been in the same school for at least3 years. Blasquez (1998) used both quantitativeand qualitative methods to see the extent to whichteachers’ understanding of mathematical conceptsinfluenced their ability to use more constructivepedagogies. She used a set of structured tasks toassess their understanding of mathematical con-cepts and used a short questionnaire to learn theirbeliefs about procedural versus constructivistapproaches to teaching. In her quantitative analy-sis, Blasquez found that teachers’ practices weremore strongly associated with their pedagogicalbeliefs than with their mathematical knowledge.For the qualitative portion of the study, Blasquezselected three pairs of teachers matched in theirbeliefs about constructivist versus proceduralinstruction, but differing in their understanding ofmathematical concepts, and then observed theirlessons. Within each pair, Blasquez was able tosee that the teacher with higher conceptual under-standing was more likely to ask students to justifytheir ideas, more likely to extend or elaborate onstudent ideas, and more likely to monitor forunderstanding. So her qualitative analysis allowedher to see an influence that her quantitative workcould not reveal. At the same time, however, theauthor was no longer “blind” for this portion ofthe study and knew which teacher belonged ineach knowledge and belief category, so the thingsshe noticed and wrote about could have beeninfluenced by her own confirmation bias.

The second study (Buckreis, 1999) was awithin-teacher comparison of a fourth-grade

teacher, Meg, selected because she had experi-ence with the fourth-grade curriculum but alsobecause her understanding of one mathematicalconcept, multiplication, was much stronger thanher knowledge of another concept, division.Buckreis evaluated Meg’s knowledge by askingMeg to list subtopics that fell within each ofthese two general curricular domains and to pro-vide sketches or diagrams as needed to showhow the topics related to one another. Meg’sknowledge of division was both faulty andincomplete on several topics, including the dif-ferent meanings of division, the conceptualunderpinnings of division procedures, and theidea of divisibility itself. The difference betweenher understanding of multiplication and of divi-sion made Meg an ideal candidate to study theinfluence of subject matter knowledge on teach-ing. In his observations, Buckreis saw that Megdid not provide students with a complete devel-opment of the full range of division situations.Moreover, at the conclusion of the observations,in a posttest, Buckreis found that students hadsignificantly more success with multiplicationproblems than with division problems.

In the third study, Stein et al. (1990) studied afifth-grade teacher, Mr. Gene, as he taught a25-lesson unit on functions and graphing. Toascertain the teacher’s knowledge of this content,these authors relied on an interview and a card-sort task, which was also administered to a math-ematics educator, so that the teachers’ knowl-edge could be defined relative to a presumedexpert’s knowledge. The contrast allows them tosee that Mr. Gene’s conception of functions wasmissing important ideas about functional rela-tionships and that much of his thinking was rel-atively superficial. For instance, in the card sort,Mr. Gene sorted cards by their format (equationsvs. graphs, for instance), whereas the mathemat-ics educator sorted them according to the natureof the relationship that they portrayed, regardlessof the format used to portray the relationship.However, the authors also suspected that theteacher’s understanding of this content hadbecome somewhat distorted by a metaphor headopted from his textbook. Mr. Gene’s textbookfrequently presented problems based on ametaphor of a function machine that producedvalues of Y for each value of X it received.Students would be asked to predict outputs for

352

Kennedy



353

different inputs. The problems also focused onpoint estimation more than on functional rela-tionships as a whole. When plotting orderedpairs, Mr. Gene emphasized the benefit of graphsto check for errors in predicting individual val-ues and gave less attention to the nature of thefunction itself. Although the authors thoughtthat Mr. Gene’s representation of functions wasinfluenced by the textbook metaphor, they alsoargued that his less sophisticated understanding

of functions led to several missed teachingopportunities and led him to view the value ofthis unit in terms of strengthening arithmeticknowledge, rather than providing a groundworkfor future content learning.

In a sense, these three studies all test a rela-tively commonsense hypothesis that “you can’tteach what you don’t know.” However, they alsoprovide a more nuanced notion of what contentknowledge actually looks like. In each case, the

TABLE 2Studies of Assessed Content Knowledge


Blasquez 3 Elementary Structure Constructivist Between-teacher Visible influence (1998) mathematics of math implementation comparison compromised

knowledge, of a prescribed by possible beliefs about lesson, student confirmation pedagogy achievement bias, hindsight

Buckreis 1 Elementary Subject matter Classroom Within-teacher Visible influence(1999) mathematics structure teaching, student comparison,

learning of multiplication multiplication vs. divisionand division

Stein, Baxter, 1 Elementary Knowledge of Representation of Thick description Visible influence & Leinhardt mathematics functions functions and compromised (1990) and graphs graphs by possible

hindsight,confirmation bias

Smith (1989) 1 Elementary Six domains of Planning and Thick description, Visible influencescience knowledge teaching teacher

science lessons attribution of learning

Magnusson, 6 Middle school Conceptual Change in Between-teacher No visible Borko, science knowledge, pupil comparison influenceKrajcik, & pedagogical knowledgeLayman content (1992) knowledge

Cunningham 6 Secondary Sociological Pedagogy used Between-teacher Visible influence (1995) science understanding in innovative comparison compromised

of science curriculum unit by possible confirmation bias

Gess- 5 Secondary Subject matter Unit planning Between-teacher Weak influenceNewsome science structure mostly, some comparison,(1992) other aspects teacher

of practice attribution of learning

Lederman 18 Secondary Nature of Ethnographic Between-teacher No visible & Zeidler science Scientific observation comparison influence(1987) Knowledge

Scale



missing knowledge was not discrete facts but,rather, seemed to be the structural relationsamong concepts. At the same time, two studieswere susceptible to confirmation bias. Moreover,their references to what the teacher failed to sayto their students are likely influenced by hind-sight. The Buckreis study overcame both ofthese potential weaknesses by augmentingobservation data with student test data.

All three studies could have been strength-ened if they had more explicitly sought evidenceregarding alternative explanations for whatteachers failed to say. What if, for instance,Mr. Gene had been interviewed after a lesson inwhich he failed to say something, and he volun-teered in the interview that in retrospect he real-ized he should have mentioned this issue? Inthis case, we would have to attribute the failureto speak as caused by something other thanmissing content knowledge.

Elementary and middle school science. Twostudies examined the role of content knowledgein the teaching practices of elementary or middleschool science teachers. The first study (Smith,1989) described a single fifth-grade teacher asshe taught a unit on light and shadow. Smith wasspecifically interested in the teachers’ knowledgeof light and shadow but she also examinedknowledge of classroom organization and man-agement, children’s thinking, science teachingstrategies, and curriculum along with her orien-tation to teaching science. Smith combinedvideo-stimulated interviews with a teacher jour-nal and close analyses of the lessons to showwhere and how each of these six kinds of knowl-edge influences the teacher’s practice. Forinstance, she described an occasion in which stu-dents were creating shadows of different objectsby holding them in front of a light. One studentsaid she could make the shadow of a piece ofhair disappear by holding the hair close to thelight. This student’s discovery contradicted thelesson’s messages that even the smallest objectsblock light and therefore create shadows. Theteacher was able to spontaneously account forthis apparent contradiction by reminding stu-dents that when objects are closer to the light,they make larger shadows, and when a hair isplaced so near the light, its shadow is too largeand diffuse to be visible. She then demonstrated

that when the hair is pulled further away fromthe light, its shadow was visible. The context ofthis study—a professional development programand a researcher who was part of it—invites rivalhypotheses about observer confirmation bias andpossibly a teacher trying to accommodate theresearcher’s values; however, the sheer volumeof evidence presented, and the details provided,is persuasive enough to override these alternativeexplanations.

The second study (Magnusson, Borko,Krajcik, & Layman, 1992) examined the rela-tionship between teacher knowledge and gainsin student content knowledge in the context of acomputer-based laboratory unit on heat energyand temperature. The study used a logic that ismore akin to quantitative studies than to qualita-tive: Researchers did not observe classroompractice at all but focused entirely on qualitativeassessments of both teacher knowledge and stu-dent knowledge before and after the unit. Theknowledge of six eighth-grade teachers and asample of their students was assessed via struc-tured tasks and interviews whose transcriptswere coded for correct or incorrect statementsand for misconceptions. The authors then pres-ent complex tables showing patterns of whatteachers knew and what their students learned.For example, they show that when teachersoffered incorrect knowledge or misconceptionson the structured tasks, their students learnedless during the year. And when teachers lackedknowledge of the misconceptions their studentsmight have, their students learned less. However,the single greatest predictor of student learningwas not what teachers knew, according to thisassessment, but rather the number and type oflearning activities they said they provided fortheir students. Although it may be the case thatthese learning activities derived from teachers’knowledge, they apparently did not derive fromthe knowledge that the researchers assessed. Noclear path between knowledge and practice wasestablished.

These two science studies offer an interestingcontrast because they provide such differentapproaches to evidence and inference. Smithinundated us with examples of specific class-room events, providing a two-volume study of asingle teacher, whereas Magnuson and col-leagues provided only pre- and postinterviews

354

Kennedy



with no observations. Smith’s study shows uswhere and how Meg’s knowledge influences herteaching practice, whereas Magnuson and col-leagues show us a pattern of regularities thatsuggest student learning is more influenced bylearning activities than by the teachers’ knowl-edge. Any number of interpretations can beoffered for the Magnusson data—that the teach-ers’ knowledge really is not relevant, that teach-ers with greater or better knowledge were morelikely to spend time teaching these topics, thatteachers with greater insight into their students’comprehension were more likely to teach theadditional units, and so on. A more in-depthapproach to qualitative work might have pro-vided insights into these issues.

Secondary science. Three studies focused onsecondary science content knowledge. In thefirst, Cunningham (1995) studied teachers’understanding of the sociology of science—thatis, how scientists interact with each other tocreate new knowledge. Thinking that a betterunderstanding of these processes might lead toa more progressive approach to teaching, theauthor assessed teachers’ knowledge of the soci-ology of science, selected six science teacherswho varied in their knowledge of this domain,and then observed them as they implementedtwo specific curricular units. All teachers hadparticipated in a professional development pro-gram designed to give teachers more under-standing of how social processes influenceresearch production and to encourage teachersto use more group processes in their own sci-ence classes. Cunningham found that teacherswith stronger sociological understanding usedmore group work in their classrooms and thatthey acknowledged a greater variety of sourcesof valid scientific information. Because allteachers had participated in the same program,accommodation bias should not be an issue.However, Cunningham’s findings might still besusceptible to confirmation bias, for she couldhave noticed more “good” practices whenteachers had more knowledge and more “not-so-good” practices in other teachers’ classrooms.This hypothesis is made more likely by the factthat Cunningham stipulated at the outset thatsociological understanding of science shouldinfluence teaching practice.

In the second study, Gess-Newsome (1992)examined “subject matter structure” (SMS),referring both to the central concepts and themesin biology and to how different specific topicsrelate to one another and to these central con-cepts. Focusing on five secondary biology teach-ers, Gess-Newsome deciphered the SMSes thatwere implied in their textbooks, the SMSes thatwere conveyed through teachers’ classroom pre-sentations and discussions, and the SMSes thatteachers revealed in a postinterview about therelationship among curricular units and theteachers’ rationale for a variety of other teachingdecisions. Analyses of classroom lessonsfocused on explicated relationships among cur-ricular topics and on explicit themes. Ultimately,the author found that these observed SMSeswere not strongly related to teachers’ own mapsof the content. Only one teacher demonstrated adirect translation of his own understanding of thematerial into his instruction. The remaining fourdiffered in their desire to provide such an organ-izing framework to their students. However, evenwhen this was not their explicit goal, teacherswith greater knowledge of relations among top-ics did spontaneously make integrative connec-tions and provided real-life examples duringtheir classroom lessons.

The third study of secondary science teacherknowledge focused on teachers’ understandingof the nature of scientific knowledge (NSK;Lederman & Zeidler, 1987). These authorsrecruited 18 experienced biology teachers andassessed their knowledge with a set of Likert-type scales asking teachers to agree or disagreewith statements describing scientific knowl-edge. Items were then aggregated into subscalesmeasuring teachers’ perceptions of the extent towhich scientific knowledge is amoral, parsimo-nious, testable, tentative, creative, and parsimo-nious. Classroom observers had no knowledgeof teachers’ scores on the NSK scales, thus pro-tecting themselves from possible confirmationbias. Ultimately, these authors used quantitativeprocedures to test for relationships betweenteachers’ knowledge and their practices. Theyranked teachers according to their scores regard-ing the nature of scientific knowledge andthen looked at the patterns in the messages thatdifferent teachers conveyed to their studentsabout these issues. From the classroom data,

355




they generated 44 categories of messages butfound only 1 of them to be statistically signifi-cantly related to teachers’ NSK scores. Becausestatistical significance typically means there is1 chance in 20 that the finding is because ofchance alone, and because 44 tests were madeby these authors, chance alone would lead us toexpect at least two “significant” links in thisset of data, thus suggesting that their measure ofteacher knowledge was not related to any of theirmeasures of classroom practice.

These three studies offer an interestingcomparison in part because they focus on suchdifferent aspects of scientific knowledge and inpart because they use such different approachesto ascertaining influence. Cunningham (1995)looks at the sociology of science, Gess-Newsome(1992) at how knowledge is organized, andLederman and Zeidler (1987) at characteristicsof scientific knowledge. The presence of suchdiverse approaches to defining and assessing con-tent knowledge illustrates an advantage of study-ing teachers’ knowledge qualitatively, for qualita-tive approaches allow more nuanced notions ofknowledge. Yet as a set, the findings do not pro-vide a strong case. Cunningham found an effectof sociological understanding, but her finding issusceptible to confirmation bias, and the othertwo authors found little or no influence.

Summary of studies focusing on content knowl-edge. Two groups of studies focused on teach-ers’ content knowledge, one relying on stipu-lated differences in knowledge and the otherinvolving direct assessments of knowledge. Toreconcile these groups of studies with quantita-tive literature, I ask two questions. First, are theconclusions generally comparable? And second,do the qualitative studies live up to their prom-ise and offer deeper or more nuanced under-standing of the relationship, in this case,between teachers’ content knowledge and theirclassroom practices?

With respect to the first question, the findingsare uneven across the set of studies. Three of 11studies showed a clear and unambiguous influ-ence whose validity was not threatened in anobvious way. Two others found an unambiguouslack of influence, and the rest either found weakinfluence or were compromised by threats totheir validity. That these findings are as uneven as

those of their quantitative brethren suggests thatthe problem of establishing this relationship maynot be a methodological problem. However, onedifference in study methodology is worth noting.The two studies that saw no visible influencewere actually hybrid qualitative–quantitativestudies. Both started with qualitative data butthen coded and quantified it to a point where theycould present tabular findings. It is possible thatthese efforts to standardize their data resulted inmasking the very things the authors sought.

So how do these studies add to our under-standing of the role of teachers’ content knowl-edge? They do offer a variety of insights andhypotheses for further investigation. For exam-ple, Stein et al. (1990) thought their teacher’sknowledge of functions and graphs had beeninfluenced by the function machine analogy histextbook used as a pedagogical tool, thus hint-ing that practice may influence knowledge evenas we seek evidence that knowledge is influenc-ing practice. Other authors, such as Blasquez(1998) and Gess-Newsome (1992), found thatteachers’ beliefs and goals for teaching contenthad greater influences on practice than didknowledge, a possibility that also raises ques-tions about how knowledge, beliefs, and valuesmay influence one another. Both of theseinsights also remind us that knowledge is not afixed resource that teachers repeatedly drawfrom over time, as chalk is, but rather is con-stantly changing in response to context, experi-ence, beliefs, and values. Perhaps the very ques-tion of how knowledge influences practice is illconceived. It assumes knowledge is a relativelyfixed entity that can be called on at will andoverlooks the role of goals, beliefs, and sponta-neous interpretations of events in determiningwhat teachers do at any given moment.

Influence of Specific Teacher Education Courses

In addition to content knowledge, researchersand policy makers have been concerned aboutthe value of that portion of teachers’ educationthat is typically labeled “teacher education.”Few quantitative studies address the merits ofparticular content within teacher education.Instead, they tend to focus on general categoriesof experiences, such as college majors or

356

Kennedy



357

advanced degrees. One recent study examinescourse-taking patterns for all teachers in Florida(Harris & Sass, 2006) and finds no evidence thatpreparation in teacher education itself con-tributes to student learning. This section reviewsthree studies of specific courses within teachereducation programs and the section belowreviews studies of whole programs or programsegments. One advantage all of these studieshave over quantitative studies is that they focuson people who have recently graduated fromteacher education and are in that sense compa-rable. Quantitative studies often include wholepopulations of teachers, including those whograduated in the past year and those who gradu-ated 30 years ago, so we are uncertain whatexactly their education might have been like,and we are uncertain as to whether it is even rea-sonable to ask about influences of programsacross such broad time horizons.

Three studies examined the influence of spe-cific teacher education courses. One (Boedecker,1997) examined the influence of science meth-ods courses; the other two (Artiles et al., 1998;Causey, Thomas, & Armeto, 2000) examined theinfluence of courses that were designed to fostercultural sensitivity in their students. All three

studies followed students from their teacher edu-cation programs into their first teaching jobs andlooked for evidence of course teachings in eitherthe teachers’ practices or their reasoning abouttheir practices. The studies are summarized inTable 3.

Boedecker (1997) was interested in a sec-ondary science methods course that emphasizedthe use of hands-on and inquiry-based pedago-gies. Her study focused on three graduates ofher own teacher education program, interview-ing teachers directly about program influencesand also observing them each four times.Boedecker saw some examples of constructivistapproaches but more often saw lectures. In theirinterviews, teachers attributed their pedagogyto the science courses they had taken ratherthan their science methods courses. Boedeckerbelieved that the science methods courses weremore consistent with National Science EducationStandards and speculated that teachers would bemore able to implement these standards if theirscience teachers had used them.

The next two studies were interested in theeffect of courses designed to alter cultural atti-tudes. In one study, Artiles and others (1998)followed two novice bilingual teachers who had

TABLE 3Studies of the Influence of Specific Teacher Education Courses


Boedecker 3 Secondary Science Observed practice Follow-up, Weak influence (1997) science methods and lesson plans content despite

courses match, teacher possible attribution observer bias,

accommodation bias

Artiles, 2 Elementary Course in Integration of Longitudinal Weak influence Barreto, bilingual multicultural knowledge despite Pena, & education and beliefs, possible McClafferty translation accommodation (1998) into practice bias, observer

biasCausey, 2 Middle Course in Beliefs and Follow-up, Weak influence

Thomas, school student values, content despite & Armeto social diversity pedagogy match possible (2000) studies accommodation

bias, observer bias, confounded context



been students in a course that the authors them-selves taught. The students were followedthrough 2 years of full-time practice. Theseauthors conducted in-depth interviews and stim-ulated-recall interviews and also asked teachersto create concept maps on several occasions.The article presents great detail about the con-cept maps and how they changed over time.From these multiple sources of evidence, theauthors concluded that some important transfor-mations had occurred in their thinking over timebut not in all domains. For example, teachersadded knowledge details to their maps but oftendid not change superordinate beliefs, and theirrationale for teaching decisions often did notmatch self-reported pedagogical beliefs. Theauthors concluded that the teacher educationprogram gave them general ideas but not thetools they needed to cope with the demands of aculturally diverse classroom.

Finally, Causey et al. (2000) were interestedin a middle school social studies methods coursedesigned to foster a more culturally sensitiveattitude. They identified a set of beliefs thatentering students tended to hold and that theyhoped to alter. In particular, they felt students didnot appreciate their own privileged positions. Intheir study, they examined student reflectionsthroughout the course and found that theirbeliefs did not change much. However, two stu-dents did appear to have genuinely changed, andso they followed these two students after theygraduated. The follow-up consisted of oneobservation and a stimulated-recall intervieweach year. The authors found that one of the twostudents had reverted to her former, less cultur-ally sensitive beliefs, but the other appeared tohave sustained the changes she had made duringthe class. However, this teacher worked in a pre-dominantly White, middle-class suburb—whereher new beliefs may not have been challenged—whereas the teacher who reverted taught in amore diverse, lower-middle-class school.Ultimately, then, the course had no discernableeffect on the practices of urban teachers.

Influence of Whole Teacher Education Programs

The value of the teacher education programas a whole is the most contentious part of

debates about teachers’ qualifications. Althoughstate requirements for certification are remark-ably diverse, all require teachers to obtain abachelor’s degree and to take some number ofcourses in teacher education. Surprisingly, veryfew quantitative studies have examined the valueof teacher education programs per se and insteadoften limit their “teacher education” question towhether teachers are certified or whether theyhave obtained advanced degrees, even thoughrequirements for certification and degreesinclude many features other than teacher educa-tion courses per se. Many literature reviews (e.g.,R. Greenwald et al., 1996; Wayne & Youngs,2003) also focus on credentials rather than onprogram characteristics. Only one (Zeichner &Conklin, 2005) reviews studies that contrastalternative approaches to teacher educationitself. Zeichner and Conklin (2005) concluded,from their review, that the literature is too messyto yield any coherent conclusions. They werealso disappointed that so much of the literatureaddressed structural differences (e.g., 4-yearrather than 5-year programs) rather than substan-tive differences. This is an area, then, wherequalitative research may shed light. All of thestudies reviewed here examined the content actu-ally taught to students and looked for evidencethat this specific content influenced teachersafter they took full-time teaching positions.

Just as it would be a mistake to assume thatteachers’ education consisted solely of their spe-cific teacher education program, so would it bea mistake to assume that any of these studiesexamines the impact of an entire educationalprogram. Most study one aspect of the program,such as the science preparation program or theliteracy preparation program. Nine studies aredescribed here and are summarized in Table 4.These studies were quite various, focusing onelementary literacy, mathematics, and urbaneducation and on secondary science, English,and physical education.

Elementary reading and literacy. Two studiesfollowed teachers who had studied in elementaryliteracy programs. One study, the BeginningTeacher Study (Flint et al., 2001; Hoffman et al.,2005; Maloch et al., 2003) was largely quan-titative (described in Hoffman et al., 2005) but included some qualitative reports as well

358

Kennedy


guest.venture

Elementary


359

TABLE 4Studies of the Influence of Specific Teacher Education Programs


Flint et al. 42, Elementary Program in Practices that Between- Visible influence (2001), 101 reading reading vs. reflect program teacher compromised Maloch elementary concepts comparison by possible et al. accommodation (2003) bias

Deal & 2 Elementary Constructivist Program influence Longitudinal, No visible White literacy program on teacher content influence but (2005, decisions match, potential 2006) teacher reverse

attribution confirmation bias

Cady 2 Elementary Program in Change in Longitudinal Influence (2006a, mathematics math beliefs and compromised 2006b) education in practice by potential

professional development program

Schultz, 4 Elementary Program in Specific Follow-up Weak influence Jones- urban urban pedagogy despite potential Walker, & education accommodation Chikkatur bias, confirmation (2008) bias

Adams & 4 Secondary Program in Evidence that Follow-up, Weak influence Krockover science science teachers content despite possible (1997) education translated match, accommodation

concepts into teacher bias,practice attribution misspecified

influenceGrossman 6 Secondary Added value Representation of Between-teacher Visible influence

(1988a, English of teacher content in comparison, compromised 1988b, education classroom teacher by possible 1989, after a strong lessons attribution observer bias,1990), liberal arts of learning accommodation Grossman education bias,& Richert misspecified (1988) influence

Grossman 3 Secondary Program in Evidence that Longitudinal Weak influence et al. English, English teachers study, despite (2000) language education appropriated content possible

arts concepts match, other observer biasNolan 2 Physical Physical Specific Follow-up, No visible (1995) education, education pedagogy content influence

any level teacher match despite education potential

accommodation bias

Valli & 1 Secondary Teacher Classroom Longitudinal Visible influence Agostinelli mathematics education control, study, compromised (1993) vs. not planning, before and by potential

teaching after observer bias,practices confirmation

bias, reactivity,confounded context



(described in Flint et al., 2001, and Maloch et al.,2003). For the qualitative component, researchersconducted structured telephone interviews with1st-year teachers who had graduated from severalprograms identified as exemplary approaches toliteracy teaching. They also interviewed teacherswho had earned generic elementary credentialsat the same institutions. The researchers wantedto learn if teachers’ self-reported practices dif-fered as a function of the particular institutionthey attended or the type of program within theinstitution. The study included more teachersthan most qualitative studies (42 in Flint et al.,2001; 101 in Maloch et al., 2003) but much lessintense data collection. No face-to-face interviewswere conducted nor any direct observation ofpractice. However the interviews were repeatedthree times during the course of the teachers’ 1styear of full-time teaching. Analyses of themessuggested that graduates from exemplary pro-grams felt more confident in their ability to teachreading, were more likely to mention specificfeatures of their programs that they valued, andwere more likely to base instructional decisionson student needs rather than curricular mandates.Because these findings derive entirely from self-reported practices, they are highly susceptible toaccommodation bias. All interviewees under-stood that they were being asked by representa-tives of their alma maters to evaluate the qualityof their preparation programs.

The second study followed two teachers lon-gitudinally and focused on the interactionbetween program influences, teachers’ beliefs,and teachers’ practices. Deal and White (2006)describe the evolution of teacher thought frompreservice student teaching through the 1st yearof professional practice, and Deal and White(2005) describe the 2nd year of professionalpractice. These researchers begin with theassumption that teachers’ a priori beliefs have agreat deal of influence on practice and so focustheir study on whether or how the teacher edu-cation program influenced teaching decisionsrelative to school context and personal beliefs.They interviewed teachers about influences ontheir decisions and then sorted these into maincategories, for example, students, institutionalrules or norms, program ideas, personal beliefs,and so on. During student teaching and 1st yearof teaching, teachers made numerous references

to their teacher education program as an influ-ence, but these references virtually disappearedby the 2nd year of professional practice, anddecisions seemed to be dominated by contextualinfluences. However, the program emphasizedthe importance of attending closely to studentthinking, and in the authors’ analysis, studentswere categorized as part of the school context.Thus if teachers said their decisions were influ-enced by their students, the researchers catego-rized the influence as related to context ratherthan to the program. It is unfortunate that theydid not present more specific examples of howteachers’ perceptions of their children influ-enced their decision making so that potentialprogram influences might be more apparent.Ironically, this apparent lack of program influ-ence may be a case of confirmation bias in thesense that the researchers expected to findstronger influences from context and priorbeliefs than for their program.

Elementary mathematics. One study followedteachers over time from a program that empha-sized cognitively guided instruction (Carpenter,Fennema, Peterson, Chiang, & Loef, 1989;Fennema, Franke, Carpenter, & Carey, 1993;Vacc & Bright, 1999). The authors followed 12teachers from their student teaching through their2nd year of professional practice using a series ofsuccessive “snapshots,” each of which includedquestionnaire data as well as interviews andobservations. They then returned for a final snap-shot as teachers began their 6th year of profes-sional practice. These authors assumed thatteachers’ beliefs would be powerful mediators ofprogram messages and used formal measures toassess both teachers’ general epistemologicalbeliefs and their beliefs about mathematics teach-ing and learning. The questionnaire data wereanalyzed quantitatively and reported in Cady,Meier, and Lubinski (2006a). Two teachers werethen identified whose patterns of change werestrikingly different from one another, and theseteachers were the focus of a qualitative studyreported in Cady et al. (2006b). The authorsfound that the teacher whose beliefs were leastaligned with the program at the outset had actu-ally adopted program-recommended practiceswhile participating in the program but laterreverted to traditional practice. The teacher

360

Kennedy



whose epistemic beliefs were more aligned withthe program was slower to change but sustainedher changes over time. However, this teacher alsoparticipated in a professional development pro-gram that shared the same goals as the preserviceprogram, so it is difficult to make a case that theprogram had a clear influence on her practice.

Elementary urban education. One study (Schultz,Jones-Walker, & Chikkatur, 2008) followedteachers after they completed a preservice pro-gram that was explicitly designed to prepare themto teach in urban, high-poverty schools. The pro-gram advocated a “listening stance” to help theirpredominantly White, middle-class novice teach-ers learn to teach to students who were quite dif-ferent from themselves. The study took place in aschool with high teacher turnover and a rigid cur-riculum with a 6-week testing schedule, so teach-ers had few opportunities to tailor their curriculumto student interests. The authors focused on fourprogram graduates teaching in this school, eachof whom had 1 to 4 years of experience. Theydescribe these teachers’ practices as deriving froma negotiation between the demands of the situa-tion and their own values and beliefs. Teachersvaried in how strongly they embraced the listen-ing stance but generally found few ways to incor-porate what they learned by listening into theircurriculum. The authors attributed this difficultyin part to curricular constraints, but it is not appar-ent that the program offered any guidance on whatto do with knowledge about students that wasobtained by listening. The program may havebeen oriented more toward listening than towardincorporating what is learned from listening intocurriculum and instructional decisions.

Secondary science. One study (Adams &Krockover, 1997) sought evidence of programimpact by comparing the ideas of program grad-uates with ideas that were embodied in programrhetoric. The authors examined program docu-ments and also interviewed faculty about keyideas, concepts, and beliefs that they wantedteachers to learn. They then visited four recentgraduates (with 1 to 3 years of experience) whohad also been interviewed during their studentteaching and were known to hold different per-ceptions of the relative value of their teachereducation courses. Interview questions targeted

the source of teachers’ knowledge, and theauthors used responses to trace teachers’ ideasback to specific courses. They found that someaspects of the program did translate into prac-tice and did become part of teachers’ knowl-edge. However, the program’s influence, espe-cially its constructivist ideas, was often neutral-ized by competing influences from their ownprior beliefs, their classroom experiences, andtheir school contexts.

Secondary English. Two studies looked at theeffect of secondary English programs. The firststudy was presented in a dissertation (Grossman,1988b), a book (Grossman, 1990), and three jour-nal articles (Grossman, 1988a, 1989; Grossman& Richert, 1988). The study examined six sec-ondary English teachers, three of whom had notstudied teacher education and three of whom had.The author did not follow teachers longitudinallythrough the program but did try to match teacherson their general educational background. Thestudy focused on teachers’ pedagogical contentknowledge and provided a sophisticated theoret-ical rationale for its design and focus. To ascer-tain teachers’ pedagogical content knowledge,the author used a series of structured interviewsabout literature and teaching literature, the teach-ers’perceptions of the courses they took as part oftheir preservice program, how they planned theirown courses, and their rational for a particularunit of instruction. She also observed teachersrepeatedly during a single unit of instruction. Theauthor found that the three graduates of teachereducation held similar ideas about the purposesfor teaching English and that they frequentlyattributed their ideas to their subject-specificcourse work. They also chose a wider variety ofliterature than the other three teachers. The teach-ers without professional education found it moredifficult to anticipate what students would finddifficult in English and generally had moretrouble adapting content for students of differentability levels. Though the study provides sub-stantial detail and a strong theoretical rationale, itremains susceptible to both accommodationbias and confirmation bias. In addition, the authordocumented program messages after these stu-dents had graduated, so there is a possibility thatthe program the author saw was not the same asthe program teachers saw.

361




The second study of elementary and second-ary English teachers (Grossman et al., 2000)involves a longitudinal study of 3 novice teach-ers followed from their student-teaching yearthrough their first 2 years as professional teach-ers. The authors set out specifically to addressthe problem of why teacher education programslack a stronger influence on practice. Theystudied 3 teachers selected from a larger sam-ple of 10 but do not say how either the 10 or the3 teachers were selected. Data collection con-sisted of five observations and interviews peryear. The authors found that teachers adoptedinstructional concepts, such as scaffolding, aprocess orientation to writing, and the value ofownership in writing, from their teacher educa-tion programs but that they had to go elsewhereto find the practical tools they needed to actu-ally teach. They depended heavily on districtcurriculum materials even when these were anti-thetical to the program’s concepts yet did notabandon program concepts. From these obser-vations, the authors conclude that teacher edu-cation can play an important role in helpingteachers, for instance, by providing teachers witha vision of writing instruction, but that teacherswould benefit more if programs also providedspecific instantiations of how these visionscould be realized.

Physical education. Nolan (1995) was interestedin whether graduates of her physical educationteacher education program continued to use themethods they learned in the program after theyhad been practicing for 6 to 8 years. Nolan askedtwo former students to videotape themselvesteaching and then mail back the videotapes. Theauthor still had videotaped data of these teachersfrom when they were student teaching and wasable to do a direct comparison of “before andafter” teaching. She also conducted telephoneinterviews with each teacher. She found that bothteachers still used, albeit selectively, practicesthey had learned in the program but that in gen-eral, they engaged in less instruction and moremanagement than they had during their studentteaching and that both were more concernedwith student compliance and enjoyment thanwith student learning. Both teachers indicated ininterviews that student learning was not a prior-ity for administrators, the public, students, or

even other faculty. Instead, the general press wastoward behavior management. Nolan speculatedthat these contextual influences likely motivatedthe changes in practice.

Secondary math. One study (Valli & Agostinelli,2000) examined a single secondary math teacherwho was also the second author of the researchreport. This teacher had taught before completinghis baccalaureate degree and without havingtaken any teacher education courses. He thenreturned to college, completed a teacher educa-tion program, graduated, and returned to teach-ing. The study compares the quality of his teach-ing before and after he enrolled in the teachereducation program and concludes that he hadsubstantially improved his practice. However thisfinding is compromised by numerous alternativeexplanations. For instance, the teacher taught in adifferent context (a public high school) after par-ticipating in the program than he had taught inbefore (a Catholic high school), and the observa-tions of his initial teaching were made by a stu-dent teacher, not a researcher, and reflect theimpressions of an inexperienced teaching candi-date. The study is susceptible to accommodationbias, confirmation bias, confounding of observersover time, and confounding of teaching context.One could argue that because the teacher is acoauthor of the article, the possibilities might bemitigated, but there is no reason to believe thatparticipation as a researcher would necessarilyreduce personal investment in the findings. Thesum of these problems raises doubts about thevalidity of the findings.

Summary of studies of teacher education. Thesestudies of the influence of teacher educationcourses and programs are quite various, not justin the domains they examine but also in the out-comes they seek and in their research methods.Still, as a group, they differ from those thatexamine content knowledge in that none of thestudies of teacher education found a clear, visi-ble influence on teaching practice. One possiblereason for this difference is that the types of out-comes they seek are more difficult to achieve.Most of these programs want to alter the thingsteachers do, such as using an inquiry approachto teaching science or adopting a listeningstance in an urban school. These ideas are not

362

Kennedy



very clearly defined, and even though teachereducators describe them, their students may beunable to envision them or to appreciate the spe-cific tasks needed to pull them off. Moreover,these approaches to teaching require a lot ofspontaneous mid-lesson adjustments, thingsnovices are not good at. Kennedy (1999) refersto this problem as the problem of enactment. Sojust as findings from quantitative studies arequite various and lead to different conclusions,as the quoted passages at the beginning of thisarticle suggest, so do these qualitative studiestend to yield ambiguous conclusions.

But the qualitative studies do allow us to seemore clearly why program influences yield thepatterns that they do. Several themes appearacross these studies. One is that teachers are verystrongly influenced by the specific circumstancesin which they find themselves—the students theyserve, the curriculum and other materials at theirdisposal, organizational constraints and norms intheir buildings. Another is that program influ-ences are mediated by teachers’ prior beliefs andsometimes are entirely revised by these beliefs.Indeed, these two themes are so pervasive that themore recent publications reviewed here beginwith the assumption that program influences willbe overwhelmed by these other influences. Thethird important theme is that program contenttends to lack concrete examples. Several authorsfound that teachers claimed to embrace programconcepts but that they were unable to translatethose ideas into practice.

These themes are not new to most teachereducators. Studies of student teaching con-ducted in the 1970s pointed to the strong influ-ence of local contexts as counteracting the influ-ence of teacher education programs (see, e.g.,Hoy & Rees, 1977; Ryan et al., 1979; Zeichner,1980). I eliminated studies of student teachingfrom this review, but the similarities betweenthe literature reviewed here and the literature onstudent teaching suggest that the teacher educa-tion community has been aware for severaldecades that its programs were not powerfulenough to alter school norms.

Discussion

The current spate of arguments about theknowledge needed for teaching, and about the

value of teacher education for teaching, began inearnest after the National Commission onTeaching and America’s Future (1996) releasedits report, “Teaching and America’s Future.”That report spawned a vigorous debate about theimportance of teacher education and certifica-tion (Ballou & Podgursky, 1998, 2000a, 2000b;Darling-Hammond, 2000) that continues today.The studies reviewed here are accompanied byhundreds of quantitative studies that have alsosought associations between teachers’ knowl-edge or educational backgrounds and the qualityof their practice. Each set of studies is suscepti-ble to its own alternative explanations of find-ings, so each literature is difficult to review andsummarize.

Qualitative studies offer several advantagesfor looking at these relationships. Qualitativeresearchers tend to have more detailed andnuanced assessments of teachers’ knowledge andto know more about what teachers actuallylearned in their educational programs and whenthey completed those programs. Quantitativeresearchers often have only rough indicators ofknowledge: tallies of courses taken or test scoresbut no knowledge of the content of those coursesor tests. Moreover, quantitative samples ofteninclude teachers who completed their educa-tional programs at wildly different times in thepast. That the two sets of studies yield messy andambiguous conclusions attests to the difficulty ofsorting out causal influences in a dynamic world.

It should be noted, too, that the value of qual-itative studies for questions of causal influencehas received relatively less attention, and as aresult, less attention has been given to explicitdesign decisions. I have tried to facilitateprogress along these lines by outlining some ofthe prominent alternative explanations thatresearchers need to anticipate and control.Probably the most disappointing aspect of theseresearch reports is their lack of detail about sam-pling rationale. Yet when samples were justified,their rationales contributed to our grasp of thefindings. For instance, when Causey et al. (2000)say that the two teachers they followed were theonly two who demonstrated change during theirparticipation in class, we understand that theirtransition into professional practice could repre-sent the best possible program outcome, not arandom program outcome. When Cady et al.

363




(2006b) tell us they followed teachers whose ini-tial beliefs were differentially aligned with pro-gram goals, we understand that we will be learn-ing how the program’s messages become differ-entially translated into practice by these twoteachers. But when Grossman and others (2000)say they studied 3 teachers from a sample of 10,without saying how they selected either the 10 orthe 3, we do not know if the sample represents thebest case of learning from teacher education or ifit represents some particular kind of learningfrom teacher education.

One also wonders how much more mighthave been learned from these studies if qualita-tive researchers were more self-conscious aboutcontrolling for confounding variables, docu-menting contexts and mitigating circumstances,and taking a more skeptical stance toward theirown interpretations and evidence. Fifteen of the23 studies reviewed here were compromised bydesign flaws that raised questions about theirfindings, thus limiting their potential to help usunderstand how teachers’ knowledge and cre-dentials help them teach. Notice, too, that thelimitations I have outlined are not limitationsinherent in qualitative research. They do notsuggest that qualitative research is inherentlyweaker than quantitative research. But they dodemonstrate that when qualitative researchersestablish truth-seeking goals, they need to paymore attention to their own methods.

More important than their limitations is themessages about the role of knowledge, creden-tials, and teacher education that these studiesoffer. Because they can look at practice as itunfolds, they offer observations that quantitativestudies can rarely provide. One important mes-sage that these studies provide has to do with thetremendous power of classroom and school con-texts over practice. Schools structure teachingpractice through schedules, textbooks, materials,and rules. Students also influence teaching prac-tices. They are the primary audience from whomteachers can get feedback about their work. Theproblem of situational press is not unique toteaching, of course. We all forget about admoni-tions to eat less sugar and fat when we are hun-gry and surrounded by vending machines full ofpotato chips and candy bars.

Another important message here is that studiesseeking evidence of a role for content knowledge

were more successful than those seeking evi-dence of a role for teacher education programs.This is somewhat of a surprise, for quantitativestudies have been more equivocal about the roleof content knowledge. That studies of teachereducation programs were so rarely able to revealany clear and visible influences is particularlysurprising, because most of this literature wasgenerated by people who were themselvesteacher educators who were deeply knowledge-able of their courses and programs and thus,presumably, would be more able to see influ-ences if they were there.

The third message offered by these studies isthat teacher education courses provided onlygeneral ideas or concepts, without the kind ofdetailed guidance teachers need to translatethese concepts into specific strategies. The stud-ies introduce us to a teacher with extensive fieldexperience in earth science who is unable totranslate that experience into engaging lessons,despite having completed a teacher educationprogram, and to teachers who learned to listento their students but did not know how to incor-porate what they heard into their lessons.Teaching entails translating ideas into events,and many of the teachers described here wereunable to do that. We could speculate, as someof these authors did, that teacher educationcould be more influential if it provided exam-ples of how to translate its ideas into specificevents. This hypothesis reinforces arguments forthe use of such tools as cases and hypermediato examine specific lessons, student work, orschool contexts (for a review of this work, seeGrossman, 2005).

Finally, several studies suggest that there is atwo-way interaction between knowledge andpractice. Knowledge did not remain fixed inteachers’ minds after they left college but insteadcontinued to evolve as it interacted with bothbeliefs and teaching experience itself. Mostquantitative work is based on an assumption thatthe knowledge measured by licensure tests, orinstilled by curriculum requirements, remainsunchanged over time and that it is just as relevantto a 30-year veteran as it is to a new teacher. Butif knowledge continues to evolve over time, wewould not expect to see a relationship betweencredentials or college course work in a randomsample of teachers who graduated at different

364

Kennedy



365


times in the past. So knowledge and programmessages, both intended to be used to controlevents in the classroom, are themselves influ-enced by these same events.

These studies might also provide an impor-tant stimulus for teacher educators. Theydemonstrate many reasons why program ideasmay never find their way into classroom prac-tices and in so doing may stimulate a newgeneration of hypotheses about how programscan have a greater influence, and a new genera-tion of programs that are more grounded inschool life.

References

References marked with an asterisk wereincluded in the meta-analysis.*Adams, P. E., & Krockover, G. H. (1997). Beginning

science teacher cognition and its origins in thepreservice secondary science teacher program.Journal of Research in Science Teaching, 34(6),633–653.

Ahn, S., & Choi, J. (2004, April). Teachers’ subjectmatter knowledge as a teacher qualification: ASynthesis of the quantitative literature on students’mathematics achievement. Paper presented at theannual meeting of the American EducationalResearch Association, San Diego.

*Artiles, A. J., Barreto, R. M., Pena, L., & McClafferty,K. (1998). Pathways to teacher learning in multi-cultural contexts: A longitudinal case study of twonovice bilingual teachers in urban schools. Remedialand Special Education, 19(2), 70–90.

Au, W. (2007). High stakes testing and curricularcontrol: A qualitative metasynthesis. EducationalResearcher, 36(5), 258–267.

Ball, D. L., Hill, H. C., & Bass, H. (2005). Knowingmathematics for teaching. American Educator,29(3), 14–16, 20–22, 43–46.

Ball, D. L., Lubienski, S. T., & Mewborn, D. S. (2001).Research on teaching mathematics: The unsolvedproblem of teachers’ mathematical knowledge. In V.Richardson (Ed.), Handbook of research on teach-ing (4th ed., pp. 433–456). Washington, DC:American Educational Research Association.

Ballou, D., & Podgursky, M. (1998). The case againstteacher certification. Public Interest, 132, 17–29.

Ballou, D., & Podgursky, M. (2000a). Reformingteacher preparation and licensing: Continuing thedebate. Teachers College Record, 102(1), 5–27.

Ballou, D., & Podgursky, M. (2000b). Reformingteacher preparation and licensing: What is the evi-dence? Teachers College Record. Retrieved August

30, 2004, from http://www.tcrecord.org/content.asp?contentid=10434

Becker, H. S. (1996). Epistemology of qualitativeresearch. In R. Jessor, A. Colby, & R. Schweder(Eds.), Ethnography and human development:Context and meaning in social inquiry (pp. 53–71).Chicago: University of Chicago Press.

Begle, E. G. (1972). Teacher knowledge and studentachievement in algebra. Stanford, CA: StanfordUniversity.

*Blasquez, E. (1998). The relationship betweenteachers’ conceptual framework and pedagogicalbeliefs to the structure and delivery of mathemati-cal lessons. Unpublished dissertation, Universityof San Francisco.

*Boedecker, M. E. (1997). What influences scienceteaching? A study of three novice rural scienceteachers (Teacher education, rural education,national science education standards). Unpublisheddissertation, Oklahoma State University.

Boyd, D., Lankford, H., Loeb, S., & Wykoff, J.(2003). Analyzing the determinants of the match-ing of public school teachers to jobs. RetrievedOctober 15, 2004, from http://www.nber.org/papers/w9878

*Buckreis, W. F. (1999). Elementary mathematicsteacher subject matter knowledge and its relation-ship to teaching and learning. Unpublished disser-tation, Oregon State University.

*Cady, J., Meier, S. L., & Lubinski, C. A. (2006a).Developing mathematics teachers: The transitionfrom preservice to experienced teacher. Journal ofEducational Research, 99(5), 295–305.

*Cady, J., Meier, S. L., & Lubinski, C. A. (2006b).The mathematical tale of two teachers: A longitu-dinal study relating mathematics instructionalpractices to level of intellectual development.Mathematics Education Research Journal, 18(1),3–26.

*Carlsen, W. S. (1997). Never ask a question if youdon’t know the answer: The tension in teachingbetween modeling scientific argument and main-taining law and order. Journal of ClassroomInteraction, 32(2), 14–23.

Carpenter, T. P., Fennema, E., Peterson, P. L., Chiang,C.-P., & Loef, M. (1989). Using knowledge of chil-dren’s mathematics thinking in classroom teach-ing: An experimental study. American EducationalResearch Journal, 26, 499–531.

*Causey, V., Thomas, C. D., & Armeto, B. J. (2000).Cultural diversity is basically a foreign term to me:The challenges of diversity for preservice teachereducation. Teaching and Teacher Education, 16,33–45.

*Cunningham, C. M. (1995). The effect of teachers’sociological understanding of science on classroom



366

Kennedy

practice and curriculum innovation. Unpublisheddissertation, Cornell University.

Darling-Hammond, L. (2000). Teacher quality andstudent achievement: A review of state policy evi-dence. Educational Policy Analysis Archives, 8(1).Available from http://epaa.asu.edu/epaa/v8n1/

Darling-Hammond, L., & McLaughlin, M. W.(1999). Investing in teaching as a learning profes-sion: Policy problems and prospects. In L. Darling-Hammond & G. Sykes (Eds.), Teaching as thelearning profession: Handbook of policy and prac-tice (pp. 376–411). San Francisco: Jossey-Bass.

*Deal, D., & White, C. S. (2005). Evolving influ-ences and beliefs of two novice teachers: The 3rdyear of a longitudinal study. Action in TeacherEducation, 27(2), 54–67.

*Deal, D., & White, C. S. (2006). Voices from theclassroom: Literacy beliefs and practices of twonovice elementary teachers. Journal of Research inChildhood Education, 20(4), 313–329.

Donmoyer, R. (2001). Paradigm talk reconsidered. InV. Richardson (Ed.), Handbook of research onteaching (4th ed., pp. 174–197). Washington, DC:American Educational Research Association.

Fennema, E., Franke, M. L., Carpenter, T. P., &Carey, D. A. (1993). Using children’s mathematicalknowledge in instruction. American EducationalResearch Journal, 30(3), 555–583.

*Flint, A., Leland, C. H., Patterson, B., Hoffman, J. V.,Sailors, M. W., Mast, M. A., et al. (2001). “I’m stillfiguring out how to do this teaching thing”: A cross-site analysis of reading preparation programs onbeginning teachers’ instructional practices and deci-sions. In C. M. Roller (Ed.), Learning to teach read-ing: Setting the research agenda (pp. 100–118).Newark, DE: International Reading Association.

Friedman, S. J. (2000). How much of a problem? Areply to Ingersoll’s “The Problem of UnderqualifiedTeachers in American Secondary Schools.”Educational Researcher, 29(5), 18–20.

*Gess-Newsome, J. (1992). Biology teachers’percep-tions of subject matter structure and its relation-ship to classroom practice (Teacher knowledge).Unpublished dissertation, Oregon State University.

Goldhaber, D. D., & Brewer, D. J. (1999). Teacher licen-sing and student achievement. In M. Kanstoroom& C. F. J. Finn (Eds.), Better teachers, betterschools (pp. 83–102). Washington, DC: Thomas B.Fordham Foundation.

Greenwald, R., Hedges, L., & Laine, R. D. (1996). Theeffect of school resources on student achievement.Review of Educational Research, 66(3), 361–396.

Greenwald, A. G., Pratkanis, A. R., Leippe, M. R., &Baumgardner, M. H. (1986). Under what conditionsdoes theory obstruct research progress? Psycho-logical Review, 93(2), 216–229.

*Grossman, P. L. (1988a). Learning to teach withoutteacher education. Teachers College Record, 91(2),191–208.

*Grossman, P. L. (1988b). A study in contrast:Sources of pedagogical content knowledge for sec-ondary English. Unpublished doctoral dissertation,Stanford University.

*Grossman, P. L. (1989). A study in contrast: Sourcesof pedagogical content knowledge for secondaryEnglish. Journal of Teacher Education, 40(5), 24–31.

*Grossman, P. L. (1990). The making of a teacher:Teacher knowledge and teacher education. NewYork: Teachers College Press.

Grossman, P. L. (2005). Research on pedagogicalapproaches to teacher education. In M. Cochran-Smith & K. M. Zeichner (Eds.), Studying teachereducation: The report from the AERA Panel onResearch and Teacher Education (pp. 425–476).Washington, DC: American Educational ResearchAssociation.

*Grossman, P. L., & Richert, A. E. (1988).Unacknowledged knowledge growth: A reexami-nation of the effects of teacher education. Teachingand Teacher Education, 4, 53–62.

*Grossman, P. L., Valencia, S., Evans, K., Thompson,C., Martin, S., & Place, N. (2000). Transition intoteaching: Learning to teach writing in teacher edu-cation and beyond. Journal of Literacy Research,32(4), 631–662.

Harris, D., & Sass, T. R. (2006, April). The effects ofteacher preparation on teacher value-added. Paperpresented at the annual meeting of the AmericanEducational Research Association, Chicago.

Hoy, W. K., & Rees, R. (1977). The bureaucraticsocialization of student teachers. Journal ofTeacher Education, 28(1), 23–26.

Hundersmarck, S. (2004, April). The problem of syn-thesizing qualitative data. Paper presented at theannual meeting of the American EducationalResearch Association, San Diego, CA.

Kennedy, M. M. (1999). The role of preserviceteacher education. In L. Darling-Hammond & G.Sykes (Eds.), Teaching as the learning profession:Handbook of policy and practice (pp. 54–85). SanFrancisco: Jossey-Bass.

Kennedy, M. M., Ahn, S., & Choi, J. (2008). Thevalue added by teacher education. In M. Cochran-Smith, S. Feiman-Nemser, & J. McIntyre (Eds.),Handbook of research on teacher education (pp1249–1273). New York: Macmillan.

Lankford, H., Loeb, S., & Wykoff, J. (2002). Teachersorting and the plight of urban schools: A descrip-tive analysis. Education Evaluation and PolicyAnalysis, 24(1), 37–62.

*Lederman, N. G., & Zeidler, D. L. (1987). Scienceteachers’ conceptions of the nature of science: Do



367


they really influence teaching behavior. ScienceEducation, 71(5), 721–734.

*Magnusson, S., Borko, H., Krajcik, J., & Layman, J.(1992, March). The relationship between teachercontent and pedagogical content knowledge andstudent content knowledge of heat energy and tem-perature. Paper presented at the annual meeting ofthe National Association for Research in ScienceTeaching, Boston.

*Maloch, B., Flint, A. S., Eldridge, D., Harmon, J.,Loven, R., Fine, J., et al. (2003). Understandings,beliefs, and reported decision making of first-yearteachers from different reading teacher preparationprograms. Elementary School Journal, 103(5),431–457.

Maxwell, J. A. (2004a). Causal explanation, qualita-tive research, and scientific inquiry in education.Educational Researcher, 33(2), 3–11.

Maxwell, J. A. (2004b). Using qualitative methods forcausal explanation. Field Methods, 16(3), 243–264.

National Commission on Teaching and America’sFuture. (1996). What matters most: Teaching forAmerica’s future. New York: Author.

*Nolan, C. V. (1995). Changes in physical educationteacher effectiveness as a function of experience.Unpublished dissertation, West Virginia University.

Pool, J. E., Ellett, C. E., Schiavone, S., & Carey-Lewis, C. (2001). How valid are the NationalBoard of Professional Teaching Standards Assess-ments for predicting the quality of actual classroomteaching and learning? Results of six mini case stud-ies. Journal of Personnel Evaluation in Education,15(1), 31–48.

*Powell, R. R. (1997). Teaching alike: A cross-caseanalysis of first-career and second-career begin-ning teachers’ instructional convergence. Teachingand Teacher Education, 13(3), 341–356.

Rice, J. K. (2003). Teacher quality: Understandingthe effectiveness of teacher attribute. Washington,DC: Economic Policy Institute.

*Ringstaff, C., & Sandholtz, J. H. (2002). Out-of-fieldassignments: Case studies of two beginning teachers.Teachers College Record, 104(4), 812–841.

Ryan, K., Applegate, J., Flora, V. R., Johnston, J.,Lasley, T., Mager, G., et al. (1979). “My teachereducation program? Well . . .”: First-year teachersreflect and react. Peabody Journal of Education,56, 267–271.

*Schultz, K., Jones-Walker, C. E., & Chikkatur, A. P.(2008 ). Listening to students, negotiating beliefs:Preparing teachers for urban classrooms.Curriculum Inquiry, 38(2), 155–187.

*Smith, D. C. (1989). The role of teacher knowledgein teaching conceptual change science lessons(Vol. I and II) (Science education). Unpublisheddissertation, University of Delaware.

*Stein, M. K., Baxter, J., & Leinhardt, G. (1990).Subject-matter knowledge and elementary instruc-tion: A case from functions and graphing. AmericanEducational Research Journal, 27(4), 639–663.

Strauss, R. P. (1999). Who gets hired? The case ofPennsylvania. In M. Kanstoroom & C. E. Finn Jr.(Eds.), Better teachers, better schools (pp. 103–130).Washington, DC: Thomas B. Fordham Foundation.

Vacc, N. N., & Bright, G. W. (1999). Elementary pre-service teachers’ changing beliefs and instructionaluse of children’s mathematical thinking. Journal forResearch in Mathematics Education, 30(1), 89–110.

*Valli, L., & Agostinelli, A. (2000). Teaching beforeand after professional preparation: The story of ahigh school mathematics teacher. Journal ofTeacher Education, 44(2), 107–118.

Walsh, K. (2001). Teacher education reconsidered:Stumbling for quality. Baltimore: Abell Foundation.

Wayne, A. J., & Youngs, P. (2003). Teacher character-istics and student achievement: A review. Review ofEducational Research, 73(1), 89–122.

Wilson, S. M., Floden, R. E., & Ferrini-Mundy, J.(2001). Teacher preparation research: Currentknowledge, gaps, and recommendations. Seattle,WA: Center for the Study of Teaching and Policy.

Wozniak, M. (1990). Predictors of outstanding teach-ers of the arts. Unpublished thesis, University ofWisconsin.

Yin, R. K. (1989). Case study research: Design andmethods. Beverly Hills, CA: Sage.

Zeichner, K. M. (1980). Myths and realities: Field-based experience in preservice teacher education.Journal of Teacher Education, 32(6), 45–55.

Zeichner, K. M., & Conklin, H. (2005). Teacher edu-cation programs. In M. Cochran-Smith & K. M.Zeichner (Eds.), Studying teacher education: Thereport from the AERA Panel on Research andTeacher Education (pp. 645–736). Washington,DC: American Educational Research Association.

Authors

MARY M. KENNEDY is a professor of education atMichigan State University, College of Education, 116Erickson Hall, East Lansing, MI 48824; e-mail:[email protected]. Her scholarship focuses ondefining teacher quality and on identifying the thingsthat most influence teacher quality. She has examinedthe influences of teacher education, research knowledge,attitudes and beliefs, credentials, and school context.

Manuscript received November 21, 2007Final revision received August 18, 2008

Accepted September 3, 2008



Documents

Educational Evaluation and Policy Analysismkennedy/publications/docs/Teacher... · Contributions of Qualitative Research to Research on Teacher Qualifications Mary M. Kennedy Michigan